Abstract
Most studies tracking wellbeing do not collect data across all the months in a year. This leads to error in estimating gender differences in wellbeing for three reasons. First, there are seasonal patterns in wellbeing (particularly life satisfaction and happiness) which are gendered, so failure to account for those confounds estimates of gender differences over time. Second, studies fielded in discrete parts of the year cannot extrapolate to gender differences in other parts of the year. Making inferences about trends over time is particularly problematic when a survey changes its field survey dates across years. Third, without monthly data, surveys miss big shifts in wellbeing that occur for short periods. This is a problem because women's wellbeing is more variable over short periods of time than men's wellbeing. It also bounces back faster. We show that simply splitting the data by months in a happiness equation generates a positive male coefficient in one subset of months from September to January and a negative coefficient in months February to August. Such a split has no impact on the male coefficients in an anxiety equation. Months matter.